How prediction markets work

A prediction market is a place where people trade contracts whose payoff depends on whether some real-world event happens. The contracts pay $1 if the event resolves YES, $0 if it resolves NO — so today's price tells you what the market collectively thinks the probability is.

This guide is a plain-English walk through the structure: what you're trading, how the price means what it means, who is on the other side, and why a prediction market is a different product from a sportsbook even when the surface looks similar.

What a prediction market is

The simplest version: a market for binary outcome contracts.

• Each market is a yes/no question — will the Fed cut rates by Sept 18?, will it rain in NYC tomorrow?, will candidate X win the election?
• Each market has two outcome contracts: a YES contract that pays $1 if the answer is yes, and a NO contract that pays $1 if the answer is no. (Categorical events have one named contract per outcome instead.)
• Prices fluctuate continuously between $0 and $1 as traders update their beliefs.

The platform — Polymarket, Kalshi, etc. — operates the order book or matching engine. It doesn't take the other side of your trade itself; it matches you against another trader who wants the opposite side. That's the most important structural point and everything else flows from it.

How prices reflect probabilities

Because each contract pays exactly $1 on the winning side and $0 on the other, the price at any moment tells you how the market is splitting its dollars between those two outcomes.

A YES contract trading at 47¢ means the market is willing to risk 47¢ to make 53¢ if YES — i.e. the market collectively prices the YES outcome at roughly 47% probability. NO at 53¢ is the mirror: 53% probability for NO.

YES and NO are complementary around $1.00, but the exact sum depends on which side of the book you're looking at: best asks usually sum above $1.00, best bids usually sum below $1.00, and mid-prices are often close to $1.00. Wider gaps usually reflect spread and liquidity rather than a guaranteed arbitrage.

A few specific consequences of this structure:

• You don't need to "know the odds" to read a market. The price is the implied probability — no conversion from American / decimal / fractional odds.
• Prices move continuously. As new information lands, traders re-price. A 30¢ market that becomes a 70¢ market over an hour reflects a real shift in collective belief.
• You can disagree. A trader can take the other side of the market price; whether that produces positive realized P&L depends on calibration, fees, liquidity, and outcomes over time.

For the math behind that "right more often" part — and how to size trades around it — see the Expected value (EV) guide.

Who the counterparties are

When you buy a YES contract on a prediction market, the platform matches you to another trader who is selling YES (or, equivalently, buying NO). That trader thinks the probability is lower than the price; you think it's higher; you trade.

The mechanic in slow motion:

1. You bid 47¢ for a YES contract. Your funds get reserved.
2. The seller posts 53¢ as their side. Together you're posting the full $1 the contract will eventually pay out.
3. The exchange holds the combined $1 in escrow until resolution.
4. Resolution. If YES, the exchange pays you the $1; the seller's 53¢ is gone. If NO, the exchange pays the seller $1; your 47¢ is gone.

Net P&L: you risked 47¢ to make 53¢; the seller risked 53¢ to make 47¢. Both got paid the price they thought reflected the probability.

The key thing this does not look like:

• The exchange isn't booking your bet against itself.
• The exchange has no view on the outcome and no preference about which side wins.
• There's no margin loan, no leverage, no liquidation. Maximum loss is what you put up.

How prediction markets differ from sportsbooks

This is where the surface similarity to a sportsbook stops. Both let you take a position on a yes/no question, but the products underneath are different in a few ways that matter:

A few patterns to notice:

• Where the house margin lives. A sportsbook builds a vig into the odds — Yes + No prices sum to more than $1.00, and the excess is the bookmaker's expected profit. A prediction market charges a fee on volume (or sometimes only on settlement) and keeps the YES + NO sum near $1.00; the spread between bid and ask is the visible cost-of-entry.
• Implied probability is direct. On a market, a 47¢ price is 47% implied probability. On a sportsbook, you have to convert from American/decimal odds and back out the vig before you can compare it to a probability.
• You can sell early. On a market the order book lets you exit any time at the current price. Sportsbooks sometimes offer a one-shot "cash out" but it's at their price, not the market's.
• Either side, freely. On a market you can buy YES or buy NO with equal ease — they're symmetric instruments. On a sportsbook you take the offered price; the book sets odds on each side separately.
• Counterparty risk lives in different places. On a market your funds sit with the exchange custodian (or, on-chain, with the protocol). On a sportsbook you're a creditor of the bookmaker until they settle and pay.

None of this makes one product strictly better than the other — sportsbooks are easier to learn, often have richer markets in sports specifically, and pay out predictably. But the information content of a price is different. A prediction-market price is something you can build forecasts and analytics on top of; a sportsbook line has the bookmaker's margin baked in and is harder to use that way.

Why traders use prediction markets

A handful of typical reasons:

• You have a view on a real-world event and the market's price doesn't reflect it. A prediction market lets you put money on that view directly.
• You want to hedge. A small business that depends on, say, the outcome of a regulatory decision can buy NO on the bad outcome to offset the real-world loss it would face if YES happens.
• You want to research. Market prices aggregate information from a lot of traders. Tracking how a market moves — and which pieces of news move it — is a useful signal even if you never trade.
• You want to express probabilistic conviction. Saying "I'm 60% confident in X" is a vague claim; buying YES at 50¢ for $50 on X is a precise one.

EdgeLedger is built around the third and fourth uses: it aggregates positions from supported platforms and surfaces the realized P&L behind your actual probabilistic calls.

Where to go next

• Expected value (EV) — the math behind deciding when a market price is wrong by enough to trade.
• Getting started with Polymarket — if you want to try this on an on-chain venue.
• Getting started with Kalshi — the CFTC-regulated, custodial counterpart.